Abstract:36 grouted block masonry specimens were applied to test their compressive strength, and compressive strength test results of 530 specimens in existing researches were counted. A BP neural networks model with four parameters(block compressive strength, mortar compressive strength, grouted concrete compressive strength and the ratio of grouted concrete area to gross section area of masonry) in input layer was established to predict the compressive strength of grouted block masonry.Then, based on the BP model, a simplified formula for compressive strength of grouted block masonry was deduced. In addition, the ratio(average value) of test data and calculated data of compressive strength of grouted block masonry was also analyzed.The results show that, in the statistical sample space, the compressive strength of grouted block masonry predicted by the simplified formula is suitable. The BP neural networks method is feasible in calculating the compressive strength of grouted block masonry.